62
Organometallics 1984, 3, 62-69
Syngas Reactions. 6.+ Aliphatic Alcohols and Esters from Synthesis Gas John F. Knifton," Robert A. Grigsby, Jr., and J. J. Lin Texaco Chemical Company, Austin, Texas 78761 Received July 14, 1983
A series of ruthenium bimetallic catalysts, comprising a ruthenium source, such as R u ~ ( C O )and ~ ~ ,a second metal selected from halogen-free titanium, zirconium, cobalt, manganese, and rhenium compounds dispersed in a low-melting (mp k
2
150-
5 3 8 E
100-
50-
0
2
4
6
S
10
REACTION TIME (hr.)
Figure 1. Alcohola and eaters from synthesis gas-typical reaction profile. Catalyst charge: Ru, 4.0 mmol; Co, 4.0 mmol, Bu,PBr, 10.0 g. Reaction conditions: 220 "C; 276 bar; H,/CO, 1/1.
Product profile: methanol, A;ethanol, 0; propanol, A;methyl acetate, 0 ; ethyl acetate, 0. Data point seta for each reaction time (1+10 h) illustrated here were determined in separate experiments by analysis of the total liquid + gas products. Reproducibility was checked in duplicate runs.
1. The important features of these data are as follows: (a) The predominant initial reaction is methanol formation. (b) Homologation to C&3 alcohol becomes an increasingly important competing reaction so that ethanol eventually becomes the predominant fraction in the latter stages, while total methanol content actually declines. (c) Acetate ester formation is notably slower than alcohol formation, but again the methyl ester is the initial predominant ester fraction while later it is ethyl acetate. (d) Acetic acid is present in only trace quantities under these particular operating conditions. Replotting the same data in terms of total methanol formation (including formation of all MeOH derivatives) and total carbonylation productivity leading to acetic acid and its acetate esters, as well as total homologation to higher alcohols, we find (see Figure 2) that the ratio of total methanol formations, vs. carbonylation productivity remains constant, within experimental error, throughout this experimental sequence (1:0.24 1:0.22,respectively). On the other hand, there is a substantial decrease in the ratio of total methanol to total homologation productivity (from 1:0.34 1:0.64,respectively, in Figure 2) as a function of time. Furthermore, in related experimental series, these changes in the ratio of methanol formation to homologation activity are themselves found to be sensitive to the choice of operating parameters-particularly temperature and syngas composition, as well as initial Ru and Co concentrations. A plausible explanation for these observations is that the ethanol fractions are being generated via two reaction paths-namely, methanol homologation (eq 3) and direct synthesis from CO/H2 (vide infra). Alternative explanations for these data include (a) differing rates of catalyst deactivation (Ru vs. Co) and (b) changes in reaction parameters (solvent media etc) during the course of these synthesis which affect each catalyst moiety differently. However, both batch24and continuous22Ru-
-
-
02
0
4
6
IO
8
TIME (hr.)
Figure 2. Alcohols and esters from synthesis gas: total methanol
formation, A;homologation productivity (including EtOH, PrOH, EtOAc, and PrOAc formation), 0;carbonylation productivity (including HOAc and ROAc formation),0.
c,
i
Ln.1
f
20
0
31
[LO1
-
I
2,
[R"l ![c.1
Figure 3. Alcohols and esters from synthesis gas-effect of [Ru]. (A) Catalyst charge: Co, 4.0 mmol; Ru, 0 8.0 mmol; Bu4PBr, 10.0 g (reaction conditions as per Figure 1). Product profile:
methanol, A; ethanol, 0; propanol, A; methyl acetate, 0; ethyl acetate, 0;acetic acid, X. (B) Total methanol formation, A; homologation productivity, 0;carbonylation productivity, A;as defined in Figure 2. Co recycle studies exhibit no significant problems with catalyst deactivation (MeOH productivity f 5 % upon recycle). Furthermore, the addition of even 60 mmol of methanol a t the start of experimental series, such as that depicted in Figure 1, leads to suppression of MeOH productivity only by a factor of ca. 40%. In order to determine which catalytically active species is responsible for each of these oxygenated products (Figure l),we have also examined product compositions as a function of the individual ruthenium-cobalt carbonyl derivatives present during each CO hydrogenation sequence. Catalyst Composition Studies. The importance of RU~(CO)~~-CO~(CO)~/BU~PB~ catalyst composition has been examined first as a function of varying ruthenium concentrations. Figure 3 illustrates the effect of variations in [Ru] upon C1-C3alcohol productivity and the generation of C,-C3 alkyl acetates. All other reaction parameters, including [Co], remained constant. The important con-
Aliphatic Alcohols and Esters f r o m Synthesis Gas
--
VARIABLE
Organometallics, Vol. 3, No. 1, 1984 65
[ C.]
so
-
--
60
0
-
E E
Y
I-
2
40-
a 0 a n
20
0
10
[COI 1 [RUI
20
40
-
511
31 I
Ill
1/3
I/ 5
CO/ H2
-
Figure 4. Alcohols and esters from synthesis gas-effect of [Co]. Catalyst charge: Ru, 1.0 mmol; Co, 0 4.0 mmol; Bu4PBr,10.0 g (reaction conditions as per Figure 1;designations as per Figure 3A).
Figure 5. Alcohols and esters from synthesis gas-effect of syngas composition. Catalyst charge: Ru, 4.0 mmol; Co, 4.0 mmol, Bu4PBr,10.0 g (reaction conditions as per Figure 1;(designations as per Figure 3A).
clusions drawn from an inspection of Figure 3A are as Figure 5 shows how the synthesis gas composition can follows: change the total alcohol-ester productivity and composi(a) No aliphatic oxygenates (including methanol) are tion. We find the following: formed in the absence of the ruthenium carbonyl compo(a) Carbonylation reactions-such as the formation of nent. HOAc-are favored by a CO-rich environment (e.g., where (b) Acetic acid is the predominant product when the the CO/H2 ratio is 2/1) even while the total liquid prodinitial Ru/Co molar ratios are < O h Selectivity to this acid uctivity drops off precipitously in a CO-rich environment. may reach 68 w t ?% of the liquid product under these (b) Ester formation-particularly MeOAc formation-is standard operating conditions; the product solutions show maximized with a CO/H2 ratio of close to unity. This is strong carbonyl spectra characteristic of [Co(CO),]- (vide an approximate accord with the overall stoichiometry of infra). this synthesis (eq 7). (c) Raising the Ru/Co molar ratio increases C1-C3 alcohol activity. 3CO + 4Hz CH3COOCH3 HzO (7) Maximum methanol formation is achieved with a (c) Hydrogen-rich gas by contrast tends to lower the [Ru]/[Co] ratio of ca. 1.5. Although ruthenium “melt” ruthenium catalyst thermal stability (particularly above catalysts alone are knownmto generate ethanol, maximum 240 OCZ2)and to favor hydrogenation over carbonylation yields of ethanol are seen here a t Ru/Co ratios of 1 2. activity. The product liquids thereby become richer in Similar preferred ratios have recently been reported by methanol and the ethanol/methanol mole ratio drops Doyle,28during methanol homologation studies catalyzed sharply (from 0.72 to 0.49 in Figure 5 ) as the H,/CO ratio by mixed Ru-Co clusters. is raised from 1/1 to 5/1. Replotting the same data in terms of total methanol, Solution Spectra. The typical crude product solutions homologation, and carbonylation productivity (see Figure from these triruthenium dodecacarbonyl-dicobalt octa3B) we find that while homologation activity peaks a t carbonyl/tetrabutylphosphoniumbromide catalyzed CO Ru/Co molar ratios of close to unity and methanol forhydrogenations to Ci-C3 alcohols and their acetate esters mation is initially first order in [Ru], the carbonylation display three band sets in the metal carbonyl spectral activity leading to acetic acid and its esters changes subregion (see Figure 6). The strong band a t ca. 1888 cm-’ stantially less with increased ruthenium content (this is is characteristicz9 of [Co(CO),]-. The group of bands at particularly true over the Ru/Co range 0.5-2.0). We 1955, 1990, and 2017 cm-’ is characteristic30 of the hyconclude then that acetic acid/ester generation is relatively dridoruthenium carbonyl anion, [HRU~(CO)~,]-. The third insensitive to ruthenium catalyst concentration so long as set is a pair of bands a t 2040 and 2115 cm-’ that is conthe rate of methanol formation is significantly faster than sistent3I with the presence of [(Ru(CO),Br,]-. carbonylation activity. Increasing cobalt concentrations, Changes in these spectra do occur with changes in the on the other hand, do lead to substantial improvements ruthenium-cobalt catalyst composition. In particular, in carbonylation productivity (see Figure 4). Acetic acid where catalyst solutions contain ruthenium and cobalt in is again the major product fraction a t Co/Ru ratios 11, molar ratios that equal or exceed unity, all three band sets and although C16O% selectivity in the liquid product) directly from CO/H2 by the application of ruthenium “melt” Work on new potential applications of this chemistry is continuing.
Experimental Section Triruthenium dodecacarbonyl,dicobalt octacarbonyl and related metal oxides, salts, and complexes were purchased from nificantly faster in the presence of cobalt (turnver freoutside suppliers. Tetrabutylphosphonium bromide was purquency 1.6 X low3s-l for Ru alonez0vs. 4.6 X s-l in chased from Aldrich Chemical Co. and used as received. Synthesis Figure 4), and only with the Ru-Co combination does gas was supplied by Big Three Industries in various proportions ethanol become the major product fraction (see Figure 1). of carbon monoxide and hydrogen. All high-pressure experiments Even so, under our CO hydrogenation conditions- with were conducted in 450- and 845-mL capacity Aminco pressure CO-rich syngas (Figure 5)-acetaldehyde is never more reactors constructed of 316 stainless steel, fitted with heating and than a trace product. Mixed ruthenium-cobalt carbonyls agitation means, and hooked to large, high-pressure, synthesis gas reservoirs. Each reactor was fitted with interchangeable Pyrex are now well d o c ~ m e n t e dbut , ~in~ this ~ ~ work ~ ~ ~most ~ of glass liners. the ruthenium and cobalt can be accounted for on the basis The extent of reaction and distribution of products were deof t h e species [Co(CO),]-, [HRu(CO),,]-, and termined by gas-liquid chromatography using a 6 ft X in. [Ru(CO),Br,]-. There is a t this stage no direct spectro140 to 260 “C (20 column of porous polymer programmed from scopic evidence for the formation of mixed-metal carbocm2/min He). C1-CB alcohols and their acetate ester products nyls. were isolated by fractionaldistillation in vacuo or by GLC trapping The results of 13C-labelingstudies, using small quantities and identified by one of the following techniques: GLC, FTIR, of 90% 13C-enriched MeOH at the start of the run, are NMR, and elemental analyses. illustrated in Table 11. The ethanol enrichment is only Cl-C3 Alcohol-Acetate Ester Synthesis. A mixture of a t the terminal methyl, consistent with a substantial triruthenium dodecacarbonyl (0.852 g, 4.0 mmol of Ru) and dicobalt octacarbonyl (0.684 g, 4.0 mmol of Co) dispersed in tetportion of this product coming from methanol homologarabutylphosphonium bromide (10.0 g) is transferred in a glass t i ~ (eq n ~3).~ Decreases in the ratio of methanol to holiner under N2 purge to a 450-mL capacity precursor reactor mologation productivity as a function of time (Figure 2) equipped with heating and means of agitation. The reactor is may, however, at least in part be indicative of ethanol sealed, flushed with a mixture of carbon monoxide and hydrogen generation both by homologation and direct synthesis-as (1/1 molar), and pressured to 69 bar with the same carbon outlined in Scheme I. Alkylmetal s p e ~ i e s ~are ~ *the ~ ~ v ~ monoxide-hydrogen. ~ The reactor is heated to 220 OC, with proposed common intermediates. Typical distributions rocking, the pressure raised to 276 bar by addition of the same of C1-C3 alcohol products (Figures 1 and 2) do not follow (1/1) carbon monoxide/hydrogen mixture from a large surge tank, the Schultz-Flory relation~hip,4~ but this is not surprising and the reactor held at temperature for 6 h. Pressure is maintained at ca. 276 bar by incremental additions of carbon monin view of our kinetic and spectroscopic evidence suggesting oxide/hydrogen from the surge tank. that the C1-C2 alcohol fractions are coming from different On cooling, a typical gas sample is taken and the excess gas catalytically active intermediates. is removed. The reddish brown liquid product (24.5 g) is analyzed Acetic Acid. Acetic acid is preferentially generated by GLC and Karl-Fischer titration and the following results were with the RU~(CO)~~-CO~(CO)~/BU~PB~ catalyst at high obtained:49 26.7 wt % ethanol; 14.8 wt % methanol; 9.1 wt % cobalt/ruthenium and high CO/H2 ratios (see Figures 3, n-propanol; 1.3 wt % water; 11.2 wt % methyl acetate; 17.4 wt 4,and 5, respectively), Under these preferred conditions, % ethyl acetate; 7.0 wt % propyl acetate; 6.0 wt % acetic acid. carbonylation is apparently rapid relative to hydrogenation The liquid yield increase is (24.5 - 11.5)/11.5 X 100 = 113%. activity and much of the methanol primary product is The alkanol and acetate ester product fractions are recovered converted to HOAc, as outlined in Scheme I. The [Co(Cfrom the crude liquid product by fractional distillation in vacuo. The dark red liquid residue resolidifies upon cooling. Analyses 0)4]-anion is, under these HOAc preferred conditions, of typical gas samples show the presence 0P939% hydrogen, 35% present in significant concentrations together with smaller carbon monoxide, 19% carbon dioxide, and 6% methane. quantities of HCo(CO),. Most of the cobalt charged at the beginning of these experiments can be accounted for on Acknowledgment. We wish to thank Texaco Inc. for the basis of these cobalt carbonyls, and the correlation of permission to publish this paper, M. R. Swenson, R. Figure 8 suggests it is these carbonyls or the derivatives thereof that are responsible for the carbonylation step (44) Bhasin, M. M.; Bartley, W. J.; Ellgen, P. C.; Wilson, T. P. J. leading to acetic acid formation (see Scheme I). Catal. 1978, 54, 120. 13Cdata for the enriched acetate ester byproducts (Table (45) Hwang, H. S.; Taylor, P. D., US.Patent 4 101 450, 1978, to Celanese Corp. 11) are also in accord with Scheme I. The enrichment (46) Wunder, F. A.; Arpe, H.; Leupold, E. I.; Schmidt, H. German. pattem for ethyl acetate is (a) in the case of the ethyl group Offen. 2 814 427, 1978, to Hoechst A G. ~~
~
~
(41) Foley, H. C.; Finch, W. C.; Pierpont, C. G.; Geoffroy, G. L. Organometallics 1982, 1, 1379. (42) Fahey, D. R. J . Am. Chem. SOC.1981,103, 136. (43) Henrici-Olive, G.;Oliv6, S. J . Mol. Catal. 1982, 16, 111.
(47) Kaplan, L. J . Org. Chem. 1982, 47, 5422; 1982,47, 5424. (48) Knifton, J. F.; Lin. J. J. US.Patent 4366 259, 1982, to Texaco Develop. Corp. (49) Product yield data for this experiment are as follows: MeOH, 60 mmol; EtOH 75 mmol, PrOH, 20 mmol; MeOAc, 20 mmol: EtOAc, 26 mmol; PrOAc, 9 mmol; HOAc, 13 mmol; CH,, 104 mmol.
Organometallics 1984, 3, 69-74
Gonzales, R. D. Czimskey, R. white, and D. W. White for experimental assistance and J. M. Schuster and C. L. LeBas for FTIR and NMR data. Registry No. RuOz,12036-10-1;Ru(aca&, 14284-93-6;Ru3(CO),,, 15243-33-1;Ti(aca~)~(OBu)~, 16902-59-3;Ti(OMeI4,992-
69
92-7; HzZrOz(OAc)z,14311-93-4; Co(acac)3,21679-46-9; CO, 630-08-0; coz(CO)~,10210-68-1;Rez(CO)lo,14285-68-8;MnC03, 598-62-9; Bu4PI, 3115-66-0; C7H15Ph3PBr,13423-48-8; PrOH, 71-23-8; MeOAc, 79-20-9; EtOAc, 141-78-6; PrOAc, 109-60-4; MeOH, 67-56-1; Bu4PBr,3115-68-2; AcOH, 64-19-7; EtOH, 6417-5.
Organometallic Compounds of the Lanthanides. 17.' Tris[ (tetramethylethylenediamine)lithium] Hexamethyl Derivatives of the Rare Earths Herbert Schumann, Jochen Muller, Norbert Bruncks, Harald Lauke, and Joachim Pickardt Institut fur Anorganische und Analytische Chemie, Technische Universitat Berlin, D- 1000 Berlin 12, Bundesrepublik Deutschland
H. Schwarz and Klaus Eckart Institut fur Organische Chemie, Technische Universitat Berlin, P 1000 Berlin 12, Bundesrepublik Deutschland Received June 8, 1983
The trichlorides of yttrium, lanthanum, and the lanthanides react with methyllithium in diethyl ether in the presence of tetramethylethylenediamine (tmed) to give the complexes [Li(tmed)]3[Ln(CH3)6]with Ln = Y, La, Pr, Nd, Sm, Gd, Tb, Dy, Ho, Er, Tm, Yb, and Lu. From the reaction of lutetium tri-tertbutoxide, tmed, and tert-butyllithium in pentane, [Li(tmed)z][Lu(t-C,H,),] has been obtained. The new compounds have been characterized by elemental analyses and IR and NMR spectra. The structure of [Li(tmed)]3[H~(CH3)B] has been elucidated through complete X-ray analysis. The crystals are rhombohedral R = 0.050, and with a = 1280.7 (20) pm, a = 79.84 (13)O, space group R3c, 2 = 2, D(ca1cd) = 0.96 g cmW3, 1535 observed reflections. The potential of methylating &unsaturated aldehydes and ketones has been explored for [Li(tmed)13[Pr(CH3),]and [Li(tmed)13[sm(CH3),],and it was found that 1,2 methylation is favored over 1,4 addition. Introduction The first description of an organometallic compound of a rare-earth element, the synthesis of Sc(CZH5),and Y(C2H5)3 in 1938, was an error,2 and the first attempts to prepare phenyl derivatives of lanthanum and some lanthanides by Gilman and co-workers failed. Biphenyl was the only reaction product isolated from the reactions of LaCl, with LiC6H5 in ether or from La with diphenylmercury a t elevated temperatures in a sealed tube., Wilkinson and Birmingham prepared the first organometallic compounds of the rare earths, the tricyclopentadienyl complexes of Sc, Y, La, Ce, Pr, Nd, Sm, and Gd.4 Fourteen more years were required to demonstrate the synthesis of S C ( C ~ Hthe ~ ) ~first , "homoleptic" +bonded organometallic compound of a rare-earth element.5 The reaction of phenyllithium with LaC13 or the trichlorides of some lanthanides did not give analogous derivatives. In the cases of LaC1, and PrC13, compounds of the type Li[Ln(C6H5),] were formed.6 A considerable stabilization of "homoleptic" organometallic compounds of the rare earths could be achieved (1) Part 16: Schumann, H.; Reier, F. W.; Dettlaff, M. J. Organomet. Chem., 1983, 255, 305. (2) Plets, V. M. Dokl. Acad. Nauk SSSR 1938,20, 27. (3) Gilman, H.; Jones, R. G. J. Org. Chem. 1945, 10, 505. (4) Wilkinson, G.;Birmingham, J. M. J. Am. Chem. SOC.1954, 76, 6210. (5) Hart, F. A.; Saran, M. S. J. Chem. Soc., Chem. Commun. 1968, 1614. (6) Hart, F. A.; Massey, A. G.; Saran, M. S. J.Organomet. Chem. 1970, 21, 147.
by use of bulky alkyl or aryl groups, such as tert-butyl,' 2,6-dimethylphen~l,~ neopentyl: CH2Si(CH3)3,%11 or CH[Si(CH3)3]2,10 or alkyl and aryl ligands containing built-in chelating groups, like (dimethylamino)-0-benzyl or (dimethylamin~)methylphenyl.'~J~ We have found that permethylated complexes of the rare earths can be synthesized by using tetramethylethylenediamine (tmed) as a stabilizing base.', We have undertaken single-crystal X-ray diffraction studies of two of these derivative^.'^ Infrared and lH and 13CNMR spectra of the compounds have been recorded, and some interesting features are presented.
Experimental Section AU reactions and preparations were performed by using Schlenk tubes in an atmosphere of dried, oxygen-free argon. The solvents used were dried and freed of oxygen by refluxing and keeping over NaH or potassium and distillation under argon prior to use. Anhydrous rare-earth chlorides were prepared from the pure (7) Wayda, A. L.;Evans, W. J. J. Am. Chem. SOC.1978, 100, 7119. (8)Cotton, S.A.; Hart, F. A.; Hursthouse, M. B.; Welch, A. J. J. Chem. Soc., Chem. Commun. 1972, 1225. (9) Lappert, M. F.;Pearce, R. J. Chem., SOC. Chem. Commun. 1973, 126. (10) Atwood, J. L.; Hunter, W. E.; Rogers, R. D.; Holton, J.; McMeeking, J.; Pearce, R.; Lappert, M. F. J. Chem. Soc., Chem. Commun. 1978, 140. (11) Schumann, H.; Miiller, J. J. Organomet. Chem. 1978, 146, C5. (12) Manzer, L. E.J. Organomet. Chem. 1977, 135, C6. (13) Manzer, L. E. J. Am. Chem. SOC.1978,100,8086. (14) Schumann, H.; Muller, J. Angew. Chem. 1978,90, 307; Angew. Chem., Int. Ed. Engl. 1978, 17, 276. (15) Schumann, H.; Pickardt, J.; Bruncks, N. Angew. Chem. 1981,93, 127; Angew. Chem., Int. Ed. Eng. 1981,20, 120.
0276-733318412303-0069$01.50/0 0 1984 American Chemical Society
